.

July, 1961 FVSION

P O I N T . ~ N DTEIERJrAL DECOMPOSITIOY
OF KC104 1093

of water was formed a t 703°K. from 1.11 X
mole of 2-butanol in every second. The calculated
residence time in the catalyst bed of an average
reactant molecule then becomes about one second.
Thus for every se'cond, 3.73 X 1 0 1 9 molecules of 2-
butanol are dehydrated by 3.61 X loz1sites, and
accordingly one in 100 sites is actually in use a t
any instant. Because of the excess of available
active sites, the clomplexed molecules do not inter-
fere appreciably with the approach of entering
molecules.
Acknowledgment.-This investigation was sup-
ported in part by a research grant PHS E-l354(C5)
from the Department of Health, Education, and
Welfare of the Public Health Service, and in part
by a Frederick Gardner Cottrell Grant from the
Research Corporation. The authors wish to thank
both of these sponsors.

THE FUSION POINT AND THE THERhIAL DECOMPOSITION OF

POTASSIUM PERCHLORATE
BY A. E. SIMCHEN
Scientific Department, Ministry of Defense, P.O.B. 7063, Tel Aviv (Israel)
Recezved September 18, 1960

When heated isothermally above 570", pure KC104 undergoes first a partial decomposition in the solid state, then the
resulting solid mixture melts while continuing to lose oxygen and solidifies again when the KC103 intermediately formed has
disappeared, and the KC1 content approaches about 60 mole yo; the still remaining KC104 continues to decompose until
pure KCl remains. The detailed kinetics of the decomposition is complicated because of consecutive, simultaneous and
reverse chemical reactions: (1) KC104 + KClOs + KC1; (2) simultaneously with the simple decomposition, there are the
+
reactions betxeen K(>1O4and KC1; between KC104 and KC103; (3) 4KC!03 + 3KC1O4 KCI. The initiation of the de-
composition is considered to be a monomolecular process a t germs distributed throughout the volume of the substance,
and then propagated by a branching chain of oxygen atoms migrating through the crystal lattice or through the melts.

1. General Description of Phenomena Ob- the solid salt mixture thus obtained undergoes
tained on Heating KC104.-For the fusion point further chemical reactions accompanied by the
of KC104 several different \-allies are quoted: 525- complete fusion of the sample. The fused mass,
530°,6,58811,12and 610 i l O O . 4 On the other hand, too, decomposes, its rhemical composition con-
partial decomposition on heating, giving KC1 and tinues t o change and in the long run a re-solidifica-
KClOs, has been known for more than a century.l4>l5 tion of the mass sets in at a (more or less) constant
In view of the practical importance of the thermal temperature (the exothermal effects of the chemical
decomposition of KC104 as a source of oxygen in reactions" and of crystallization being balanced by
solid rocket, propellent formulations, it seemed of the temperature-regulating mechanism of the ap-
interest t o have a clearer picture of the phenomena paratus and by the heat transfer to the surround-
in question. The vacuum decomposition appara- ings).
tJusin which the data reported below were obtained, The chemical reactions preceding3s20 and ac-
has been described in a previous paper.' companying fusion are dependent on time and
The essent'ial fact emerging from this study is temperature, and so is the heating up of the
that KC104 has no congruent fusion point. The sample and of the glass support. This leads to a
substance remaining aft'er the resolidificat'ion of timetemperature dependency of the fusion which
the fused mass is riot KC10,. When heating KC104 is given in Table I.
at a constant and sufficiently high temperature, TABLE I
and waiting for a sufficient time, it undergoes first
n partial decomposition in the solid state2 TIMESOF BEGINNING FUSIONAT DIFFERENTTEMPERA-
TURES, OF 100-hfG. SAMPLES O F PURIFIED ANALYTICAL
(1) Benrath and Braun, 2. anorg. Chem., 244, 348 (1940). KCIOI (INCLUDINGHEAT-UP TIMES)
(2) Bircumshaw and Phillips, J. Chem. Soc., 703 (1953).
( 3 ) Caban4 and BBna-d. C.R.Sc.Sc., 260, 128, 331 (1960).
Temp., Time, Mole 7' XCl in residue after isothermal
OC. min. resolidification
(4) Carnelley and Carleton-Williams, J. Chem. SOC.,37, 125 (1880).
(5) Duerre, Dissertation, Giessen, 1907. 570 120 23 3 mole yo in 1 residue; 2.2 mole % in
(6) Frankland and Dingwall, J . Chem. Soc., 61, 2747 (1887). a 2nd sample that had not melted
(7) Glasner and Simchen, Bull. soc. chim. France, 18, 233 (1951). 580 33 . . . . . .. . . . .
(8) "Gmelins Handbrich der anorganischen Chemie," 8th edn., 590 24 52.5; 52.9; 53 8
"Chlor," and fol., Verlag Chemie, 1927, p. 398.
(9) Harvey, Edmison, Jones, Seybert and Catto, J . A m . Chem. Soe., 600 20 . .. . . . . . . . .
76, 3270 (1954). 610 16 56 7
(10) Hodgman, "I!andbook of Chemistry and Physics," 41st edn.. 620 15 57.1
Chemical Rubber Publishing Co., Cleveland, Ohio. 1959, p. 627.
(11) Markowits, J . Phgs. Chem., 61, 505 (1957). The re-solidified, partially decomposed residue
112) hIarkoaitz, ihid., 62, 827 (1958). still contains appreciable quantities of KC104, but
(13) Marignac, Bibl. Univ., 46, 353 (1843).
(14) Millon, Ann. p h y s . , [3] 7, 334 (1843). almost no KC103 if one waits until the evolution of
(15) Millon, Lieh. A n 7 ~46, , 281 (1843). gaseous O2has died down. The composition of the
116) Otto and F r y , J . A m . Chem. Soc.. 46, 1138 (1923). residue is not exactly reproducible, perhaps due t o
(17) Rodgers and Wassink, Univ. of Arkansas, Final Summary R e a different extent of the decomposition in samples
port, 1 Sept. 54-31 Jan. 58. Contract No. DA-23-072-ORD-1049.
(18) J. C. Schumacher, "Perchlorates," Am. Chem. SOC.Monograph, (19) Scobai, Z. phymk. Chem., 44, 328 (1903).
Reinhold Publ. Gorp., N,aw York, N. Y..1960, pp. 37-38. (20) Simchen and Glasner, BulL SOC. chzm. France, 20, 127 (1953).
1094 A. E. SIMCHEN Vol. 65

TABLE
I1
OF ISOTHERMALLY
COMPOSITION RE-SOLIDIFIED IN MOLE% KCl
RESIDUES,
Description of the KC104 540 550 560 570 580 590 b10 til0 620°
1 Commercial I (traces of Cu) 33.3 43 3 52.7 57 7 60 0 . 60 .5 60 5 ..
52 8 60 0 61 0 62 I
2 Commercial I1 (traces of Pb) .. 31 0 .. 50 0 .. 55 0 .. 58 9 ..
3 J. '1' Baker Analyzed (traces of Fe) .. 4 4 2 2 41 0 52 3 50.2 53 9 56 7 ..
43 0 50.8
4 Recrystallized from Baker Analyzed .. .. .. 2.2 .. 52 5 .. 56 7 57.1
( N o 3) 23 3 .. <52 9 ..
53 8
KC108 who disregarded the instability of the mixture.
C In the ternary system KC10~-IiCl03-KCl, the
fnsion point of KCl a t 771°2 is the only one to be
considered as accurate; the other fusion poiiits are
t o be treated with reserve, and the same may be
said about the above mentioned eutectic tempera-
ture of KClOpKCl. If we neglect in a first ap-
proximation the thermal effect accompanying the
fusion or the decomposition in the above system,
then all changes of the chemical composition can
be regarded as the displacement of a representative
point inside the triangular diagram having the three
components KC104, KC103 and KC1 a t its summits.
For the decomposition of KC104, the representative
point will leave the KC104 corner of the diagram
and will follow a path directed toward the KCl
corner, but approaching somewhat the KClOs cor-
ner on its way (Fig. 1).
KC104 KC1 TABLE 111
Fig. 1.-Evolution
composing of the
x c 1 0 4 , ~h~ chemical
reaction pathcomposition of
is: a-d-e-b; de-
e- CHEMICAL COMPOSITION O F RESIDUESO F 100-1\1G. 8.4MPLES
point of re-solidification. The KCIOtcontent of the re- O F KClOi (RECRYSTALLIZED FROM ARALYTICAL KC10,)
solidified mass is given by e-g, the KC1-content by e-f HEATEDISOTHERVALLY FOR DIFFFRENT TIVES
(schematic). A. Heating at atm. pressure: A . l 580 f 1"
Time, min. 15 30 50 70 100 150
differently distributed in the glass spoon, differ- Mole % KC1 1 39 12 85 33 60 47 0 65 3 96 9
ences in the distribution of the grain size of the
sample before fusion, or other causes. The coni-
2 9( 8' li 6i 7 :" 3i 9""
position of the residue and the temperature of in- A.2 600 i 1'
cipient fusion are strongly influenced by the degree Time, m'n. 15 30 45 60
Mole % KCI 2 63 11 2 91 97 99 4
of impurity of the sample, and the quality of the Mole % KCIOa 68 74 98 43
impurities present. The results are given in Table Mole c7a K a o a 9 4 69 84 06 7 03 o 17
11. B. Heating a t 600 f 1" in a maintained vacuum. Initial
A tendency to a higher KC1 content in the resi- pressure: 1 p , and always less than 0.2 mm.
dues of fusion a t higher temperatures is clearly Time, min. 10 18 25 35 45 60
visible, with a limit of about 60 mole yoof I(C1; this % KCl 5 45 13 82 37 06 62 00 74 32 80 12
3Iole % Kc101 2 88 4 95 3 28 1 40 1 00 0 43
may represent the solubility of KC1 in the fused 3101e c; KCIOl 91 67 81 23 59 66 36 2 4 5o 19 45
mass. If the heating is interrupted after various
times without regard for oxygen evolution and the &4tthe point where this curve of the reaction
cooled masd submitted to chemical analysis, a path meets the solidus of the system. the fused
continuous change of chemical composition may mass will begin to solidify. The part 'IIhich is still
be noticed, as exemplified in Table 111 (see below). liquid will continue to lose oxygen at a charac-
The KC10, that first appears in the mass together teristic rate vl; the part already solidified decom-
with KCl cthe first mention of KC103 in the poses at a different and much +mer rate v2.9*o
thermal decomposition products of KC104 go back Finally the mass solidifies entirely and loses the
to 1843 ( ! ) ) , I 4 l5 disappears later leaving a mixture remaining oxygen at a rate of the order of magni-
of KC1 and KC1O4. The remaining KC104 de- tude of v 2 only, uiitil pure I(C1 remains. The time
composes entirely and pure KC1 is obtained. of re-solidification of the fused mass depends, like
The lower temperature limit for carefully purified the time of beginning fusion, upon several factors
KClO,, undergoiiig fusion accompanied by decom- such as the yuaiitity of material undergoing de-
position has been found to be about 570". This composition, and the heat transfer between the re-
temperature coincides with the eutectic tempera- acting mass and the furnace. The difference in
ture of the KC1-KC1O4 mixture reported by Ben- the decomposition rates in the solid state and in
rath and Braunl and by Harvey and co-workersg the fused state can be understood easily if the dif-
July, 1961 F U S I O N POIR’T AND
ferent modes of the transport of oxygen in both
cases are considered. In a crystal the oxygen must
make its way through the interstices and/or de-
fects of the crystal lattice, or perhaps through the
inter-mosaic fissure; even the “frozen” lattice,
ignoring the thermal vibrations, affords the space
necessary for the interstitial movement of oxygen
which may also move by jumps from anion to
anion.7 In the liquid on the other hand, there is,
in addition to the above, diffusion of oxygen, and
the movement (of gas bubbles ascending through
the liquid and giving a more efficient mode of mass
transport.
2. The Decomposition Reactions.-The de-
composition reactions of KC104 a t high temperature
have been investigated several times since 1816.21
The principal difficulty arises from the strong reac-
tivity of KC104 at high temperature and from its
sensitivity to Qualitatively, the ac-
tion of copper, iron or cobalt salts in catalyzing the
thermal decomposition of KC104 has been known
for a long time.l’
It may be that heavy metal perchlorates form
and decompose subsequently into oxides oscillating
between higher and lower states of oxidation.
Even in analytical grade KC104 heavy metal
traces are present and can be identified easily.
At the high temperatures investigated, there is a
clear diff erence between analytical KC104, and the
same product after 2-3 recrystallizations (compare
Table 11, the results at 570’). The presence of
KC103 in partially decomposed KC104 has been
reported a number of times since its first discovery
hy l\Iillon in 1843,6 l 4 l5 l Y 22 2 3 , 2 4 and contested
by others.l3 l 8 ;1 systematic investigation of
KCIOs and of KCl. as a function of heating time, in
the resolidified residues of 100-mg. samples of care-
fully purified KC104, in uacuo (less than 200 p ) and
at atmospheric pressure, gave the results repre-
qented in Table ‘[II.
The continuous loss of oxygen seems to proceed
by successive steps: KC104 --t KC103 + KC1.
The kinetics of the system are not simple: besides
the consecutive decomposition reactions mentioned
above, there are simultaneous reactions: the reac-
tion between KC1 and KC104,7*20and perhaps
others, and besides the decomposition reactions
leading to compounds poorer in oxygen, there is
the well-known ~nversereaction : 4KC101 = 3KC1-
+
O4 KC1. The fused maw decomposes with loss
of oxygen, simultaneously becoming richer in KC1.
After re-solidification of the mass the velocity of
the gas evolution diminishes to a low value and re-
mains low a? long as the temperatiire is not raised.
The higher diffusion velocity in the liquid mass is
again replwed by the slower travelling through
the crystal 1attic.e. aiid the bulk moyement of the
oxygen bubbles rising through the melt to its sur-
face and cxollecting the dissolved oxygen in their
path, ceasrs altogether.
3. The Kinetics of the Decomposition.-Previ-
ous work reported elsewhere’ 2o ha5 led Us to
coniider the thermal decomposition of KC104 as
(21) Stadion GzZhei”s dnn. 52, 2 1 3 11816).
(22) Teed, P t o c Chem Soc , 2, 141 (1886).
(23) Teed Chem Neus, 53, 56 (1886).
( 2 0 Teed, I . Cheni Soc., 61, 282 (1887).
THERMAL
DECOMPOSITION
O F KC104 1095
composed of two different basic processes. One of
them is relatively slow and corresponds to the
initiation of the decomposition, while the other
process, up to 500 times more rapid, represents
the propagation of the decomposition by a branched
chain mechanism.’ The latter is the result of the
differences between the velocity of oxygen trans-
port from a ClO,-group (x = 1 to 4), with one oxy-
gen atom possessing the necessary energy of activa-
tion, toka neighboring ClO,-group (y = 0 to 3)
on the one hand, and the velocity of activating
the ClO,-group once the favorable conditions are
obtained, ie., if a ClO,-ion poorer in oxygen is
created spatially adjacent to a ClO,-ion nhich is
richer in oxygen.
If the activation velocity is smaller than the
transport velocity, a vacant place on a C10,-
group may remain open for a time in which an
oxygen on more than one adjacent ClO,-group is
activated, and branching may result. Such a
reasoning implies the possible existence of at least
two distinct modes of oxygen transport through
the crystal lattice: (1) the diffusion of oxygen be-
tween the ions of the crystal lattice and,’or the de-
fects of the lattice, through the cracks. fissures, etc.,
of the macrocrystals; ( 2 ) the movement, by jump<.
of oxygen from the C10,-ions to neighboring C10,-
ions (x greater than y), with a time of arrest at the
different stations; (3) possibly a combination of
(1) and ( 2 ) above, part of the movement being
between the lattice constituents, and part by jiimpa
from anion to anion. Xow if the spacing of the
ions in the KC104 is sufficiently loose to permit the
transport of oxygen, it iq even looser in the molten
mass, and the “interstitial” movement of oxygen
is possible for hoth. It ieems to-day ininecessary
to postulate as we did previously,’ a rhain initia-
tion from germs on the exterior cryital iurfacc.
Such a hypothesis led to a term proportional to
the 2 / 3 power of mass in the expresion for the
velocity of gas formation, which should rather lie
replaced by a term proportional to the niasq prewit
if the initiation of decomposition at lattice defertq
is accepted; these defect5 are distributed through
the bulk of the crystals.
In the molten mass, the initiation throughout
the volume can be observed visually. If the initia-
tion of the decomposition takes place in the bulk
rather than on the surface of the crystals. then the
grain size and its diqtribution ihould be irrelevant
to the thermal decomposition iunlei- the .train*
accompanying comminution lead to come pref-
erential diFtribution of lattice defect. or diqloca-
tions; our samples were 80 to 100 me+h BSS).
Thus, instead of
dy/dt = ~ , J ( -
u y) + j i i ~- y i ’ 1 ( 1’
for initiation on the surface of the crystals, me have
dy/dt = j u ( a - v) + J , ( u - vi 2)
for initiation in the niass of the crystals ( a =
maximum quantity of oxygen gas ohtainable from
the sample up to the solidification of the melt; y =
oxygen pressure at time t ; j, and j = 1-elocity con-
stants for the initiation of the reaction aiid for the
branched chain reaction, respectively). The value
of a depends on the conditions in each experiment
1096 WEAVER,
THOMA, FRIEDMAN, HARRISAND YAKEL
INSLEY, Vol. 65

and on its fuision and re-solidification phase. An and 8 above. Rearranging eq. 8 yields
approximation may be found for a by graphically
integrating the experimental curves of gas evolu-
k(1 + h/k) = 4(d~/dt)m(l + kl/k)-' (9)
tion velocity us. time, from the beginning of the and the substitution of eq. 9 into eq. 5 gives finally
experiment to the (extrapolated) virtual cessation 4t,(d~/dt), = (1 +
ki/k) In (k/ki) = LI ( 10)
of gas evolution a t solidification. The flirther Thus from the experimentally found values of
c*nlculationsare derived from the integration of the maximum reaction velocity (dz/dt)m, and from
equation the corresponding times tm, we may calculate the
dr/dt kz(1 - 2) + ki(1 - X) (3) ratio k l / k for the different temperatures by using a
graph or a table of the function LI (eq. 10). Then
obtained by dividing both sides of eq. 2 by a; we we compute: (1) the branching chain velocity con-
have stant k; ( 2 ) the monomolecular constant kl. Ad.
2: = y/a; kl = j,; k = j a ( 4) 1. : eq. 7 gives after rearrangement
Integrating eq. 3 between 0 and t gives k = h(k/kd/[tm(l kl/k)l+ l/(t&II) (11)
and with eq. 10
k = 4(d~/dt),( 1 + ki/k) -' (12)
4.Calculation of the Velocity Constants k and A graph of the function LII may be helpful in
k , of Eq. 3.--The conditions for the velocity of solving eq. 11
oxygen evolution becoming maximum lead to
= (1 - ki/k)/2
LII = (1 + ki/k)-' X In (k/ki) (13)
X, (6) Ad 2 . : kl is found from k l / k and from k.
In order to get the time corresponding to the Values for analytical grade KC10, are contained
maximum velocity of oxygen evolution, t, eq. 6 in Table IV.
is substituted into eq. 5 giving
TABLE IV
t, = [l,'(k + kl)] X In ( k l k l ) = [k(l + k ~ / k ) ] -X~ In VELOCITYCONSTANTSk AND kl (Ea. 3) FOR ANALYTICAL
(klki) (7) KClOi
the maximum velocity of oxygen evolution is ob- Temp.,
OC. 560 570 580 590 600 610
tained by sub,dituting eq. 6 into eq. 3
10' k 4 14 5 75 5 96 10 2 13 5 15 4
(dzldt), .= k [ ( k + ki)/2k]' = k(l + ki/k)'/4 (8) 466 8 2 9 6 137 166 203
I t is useful to conserve the ratio k l / k in the eq. 7 IO4 kl 7 41 6 44 9 20 21 85 59 8 102 3

PHASE EQUILIBRIA I N THE SYSTEM LiF-YF3

F.l v E B V E R , " H . A. F R I E D M A N , I ~ H . INSLEY,'~
BYR. E. THOMA,'~C. L. A. HARRIS'^ AND H. A. YAKEL,J R . ' ~
Reactor Chemistry Division and Metallurgy Division
Oak Ridge National Laboratory,' Post Ofice Box X, Oak Ridge, Tennessee
Received September 19, 1960

The phase equilibrium diagram of the condensed system LiF-YF3 is presented. Data were obtained from thermal analysis
of heating and cooling curves and by identifying the phases present in small samples which were quenched after equilibration
a t high tempemtures. Within the system two invariant points occur, the eutectic a t 19 mole yo YF3 and 695", and the
peritectic a t 49 inole yo YF3 and 819'. The single interm$diate compound, LiF.YF:<,is formed. It melts incongruently to
YF3and liquid a t 819O, and is tetragonal, a0 = 5.26 f 0.03 A., and co = 10.94 f 0.03 A,, space group I41/a. The compound
LiF.YF3 is colorless, uniaxial (+), with refractive indices N u = 1.454, N e = 1.472. No solid solutions are formed among
the equilibrium solids in the system LiF-YF3. The melting point of YFa, 1148O, determined by other workers, was cor-
roborated in the present study. A solid state transition in YF3was shown to occur a t 1052'.

Introduction system, with the eutectic a t 18 mole % YF3 and

Recent investigations a t the Oak Ridge National
Laboratory of the fused salt mixtures of potential
use in the electrolytic reduction of Y +++ to yttrium
metal have included a review of the phase equilib-
ria in the system LiF-YF,. As a part of a study
of the complex fluoride compounds formed from
the alkitli metal fluorides and Group 111 metal
fluorides, Dergunov3 reported a phase diagram of
the system. From that investigation he con-
cluded that LiF and YF, form a simple eutectic
(1) (a) Reactor Chemistry Division. (b) Metallurgy Division
(2) Operated for the U. S. Atomic Energy Commission by the Union
Carbide Corporation.
(3) E. P. Dergunov, Doklady Akad. Nauk S.S.S.R., 60, 1185 (1948).
at 744'. The absence of intermediate compounds
appears anomalous inasmuch as examples of inter-
mediate equilibrium compounds are known4 in
many systems of LiF-MF, -R/IF2, -MF3, and/or
-MF4, where the cation rgdius of M+, M2+, &I3+
or &I4+ is as low as 0.30 A. (Be2+)and as high as
1.67 A. (Cs+). Accordingly, an investigation of
the phase equilibria in the condensed system LiF -
YF3 mas made, using the same techniques as have
been employed effectively in the determinations of
fused salt phase equilibria previously de~cribed.~
(4) E. M. Levin and H. E'. McMurdie, "Phase Diagrams for Cr-
ramists," The American Ceramic Society, Columbus, Ohio, 1959.
( 5 ) C. J. Barton, e l aZ., J . A m . Ceram. SOC.,41,63 (1958).